FBNetV5

Reported on 5 benchmarks across 5 tasks · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Methodology2 results

Neural Architecture SearchonImageNet
Top-1 Error Rate· 2021-11-19
18.2
best: 16.1 (DeepMAD-50M)
FBNetV5: Neural Architecture Search for Multiple Tasks in One Run arXiv:2111.10007
AutoMLonImageNet
Top-1 Error Rate· 2021-11-19
18.2
best: 16.1 (DeepMAD-50M)
FBNetV5: Neural Architecture Search for Multiple Tasks in One Run arXiv:2111.10007

Medical1 result

Semantic SegmentationonADE20K
Validation mIoU· 2021-11-19
40.4
best: 63.6 (ViT-P (InternImage-H))
FBNetV5: Neural Architecture Search for Multiple Tasks in One Run arXiv:2111.10007

Computer Vision1 result

Image ClassificationonImageNet
GFLOPs· 2021-11-19
0.726
best: 1478 (InternImage-H)
FBNetV5: Neural Architecture Search for Multiple Tasks in One Run arXiv:2111.10007

Audio1 result

10-shot image generationonADE20K
Validation mIoU· 2021-11-19
40.4
best: 63.6 (ViT-P (InternImage-H))
FBNetV5: Neural Architecture Search for Multiple Tasks in One Run arXiv:2111.10007